Merge pull request #6 from oduwsdl/DuplicateStoriesFix

fix duplicate error caused by removing story once it is matched

storygraph_bot/backbone.py

@@ -58,16 +58,15 @@ def mapper(overlap_threshold, cachedstories_uri_dts, stories_uri_dts, cache_stor
 
         overlap = sorted(overlap, key=lambda x: x['coeff'], reverse=True)
         maxcoeff = overlap[0]['coeff']
-        chosensgtk_story_id = overlap[0]['sgtk_story_id']
-
-        if maxcoeff >= overlap_threshold:                 
+
+        if maxcoeff >= overlap_threshold: 
+            chosensgtk_story_id = overlap[0]['sgtk_story_id']                
             new_graphs = sorted(list(stories_uri_dts[chosensgtk_story_id] - cs_uridt), reverse=True)
             matched_stories.update({story_id: {'overlap': overlap, "new_graph_timestamps": new_graphs}})
+            stories_uri_dts[chosensgtk_story_id] = set()            
         else:
             unmatched_stories.update({story_id: {'overlap': overlap}})
 
-        stories_uri_dts[chosensgtk_story_id] = set()            
-
     map_cachestories["matched_stories"] = matched_stories
     map_cachestories["unmatched_stories"] = unmatched_stories
     return(map_cachestories)

storygraph_bot/util.py

@@ -145,13 +145,12 @@ def dump_json_to_file(outfilename, dict_to_write, indent_flag=True, extra_params
 
 
 
-def get_storygraph_stories(sgbot_path, start_datetime, end_datetime):
-
+def get_storygraph_stories(sgbot_path, start_datetime, end_datetime):   
     data = {}
     try:
         cmd = (f'sgtk --pretty-print -o {sgbot_path}/tmp/current_storygraphdata.json maxgraph --cluster-stories-by="max_avg_degree" --cluster-stories --start-datetime="{start_datetime}" --end-datetime="{end_datetime}" > {sgbot_path}/tmp/console_output.log  2>&1')
         a = os.system(cmd)
-        data = get_dict_frm_file( f"{sgbot_path}/tmp/current_storygraphdata.json" )
+        data = get_dict_frm_file(f'{sgbot_path}/tmp/current_storygraphdata.json')
     except FileNotFoundError as e:
         logger.error('Please install storygraph-toolkit: https://github.com/oduwsdl/storygraph-toolkit.git')